What Muscles Do Rows Actually Work?
The back does most of the work in a row, but the rear shoulders and biceps get a real, smaller share too. Two studies on muscle activity show how grip and torso angle shift that split, and what a week of nothing but rows leaves undertrained.
A row works the back first. The lats and the muscles across your mid and upper back do most of the pulling, because that is the job a row is built around. But two muscles most people never credit get real work on every rep: the rear part of the shoulder, and the biceps, which is doing more than holding the handle steady. None of the four share the load evenly, and which row you pick changes the split more than most gym advice admits.
The back does most of the work, and "the back" means more than the lats
Pull a bar or a handle toward your body and the lats do the heavy lifting: they run from your upper arm down to your lower back and pelvis, and pulling your arm down and back is exactly what they are shaped to do. A row does not stop there. As the elbow travels past your torso, your shoulder blade moves toward your spine as part of the same motion, and the muscles across your mid and upper back that do that job come on hard right alongside the lats.
A study that measured muscle activity directly, with electrodes on the skin, put numbers on this. Across a wide-grip pulldown, a reverse-grip pulldown, and a seated row tested two ways, with the shoulder blades actively pulled together and without, the seated row produced the highest reading in the mid-back muscles of any exercise tested, and the widest gap between how hard the back worked and how hard the biceps worked (Lehman, 2004). One thing that did not move the number, in the same study: consciously squeezing the shoulder blades together at the top of the row, on top of the pull itself, made no measurable difference to the mid-back reading. The row does the work. The extra squeeze is form coaching, not extra volume.
Do rows actually train the rear shoulders?
This is the one that surprises people. The rear part of your shoulder sits behind the joint, and pulling your elbow back and past your torso is one of the few things that trains it directly. A row does that on every rep, so it earns a real share of the set, not a token one.
What it does not earn is the largest share. A 2024 systematic review pooling thirty-three studies on shoulder-muscle activity found pull-ups and the bodyweight inverted row produced the strongest rear-shoulder readings among common back exercises, and that a reverse pec-deck fly beat the seated cable row for the same muscle (Rabello et al., 2024). The seated row still trains the rear shoulder. It is just not the exercise you would pick if the rear shoulder were the whole point.
The biceps are doing more than steadying the handle
Bend your elbow against resistance and the biceps is working, whatever else is happening at the shoulder. The same measurement study found biceps activity held roughly steady across every pulling exercise it tested: wide-grip pulldown, reverse-grip pulldown, both seated-row variations. Change your grip or your row angle and the back and mid-back numbers move around. The biceps mostly does not.
That holds only while your grip lets the elbow bend properly. A grip so wide the arm barely folds turns a row into something closer to a shrug with the arms along for the ride, and the biceps share drops with it.
Grip and torso angle move the split further than people think
Underhand and overhand are not two ways of holding the same exercise. Turn your palms up and the lats sit in a stronger pulling position, and the biceps, already flexed at the wrist and forearm, take on more of the work than they do with palms facing in or down. That is the same principle behind a pull-up: turn your grip underhand and it moves noticeably closer to a biceps exercise, without becoming one.
Torso angle matters too, separately from grip. Lean forward into a bent-over row and the bar travels a longer path across more of the back. Sit upright at a machine with your chest supported and the same pull travels less far, asks less of your lower back for stability, and leans slightly harder on the arms to finish it. Neither is the "correct" row. They are different exercises that happen to share a name.
One arm at a time changes the picture again. A single-arm dumbbell row lets the working side pull through a fuller range than a two-handed bar allows, and it removes the other arm's ability to quietly do more of the work on a rep that felt harder on one side. If a two-handed row has ever felt lopsided without you being able to say why, that is usually where the answer is: not a different muscle being trained, but one side finally unable to hide behind the other.
What a week of rows and nothing else actually misses
Here is where the numbers matter for your own week, not just for interest. Most people who track weekly sets per muscle give an assisting muscle credit for roughly half a working set rather than a full one, which is the standard convention for tallying volume that trains a muscle indirectly. A row that earns the back a full set only earns the rear shoulder about half of one.
Run that forward across a week that is entirely rows and pulldowns, with no face pull, no rear-delt fly, nothing aimed at the rear shoulder on purpose. The back reaches its weekly number in the same number of sets it always would. The rear shoulder, banking half credit each time, needs roughly twice as many rowing sets to reach the same target, and most training weeks do not have room for that many. It is not that rows skip the rear shoulder. It is that closing the same gap with rows alone costs about double the sets a dedicated exercise would, which is the same arithmetic behind scoring each muscle against its own weekly target rather than lumping the whole back into one number.
Checking this against your own week
Open your training log for the last seven days and count two things: how many sets involved a row, a pulldown, or a pull-up, and how many involved anything aimed at the rear shoulder on its own, a face pull, a reverse fly, a rear-delt raise. Most people find the first number and not the second. That gap is exactly the one the numbers above predict, and it is easy to see once you know where to look in your own week rather than someone else's.
Put a number on it. Six rowing sets in a week is a full week for the back on most weekly targets. Credited at roughly half each, those same six sets add up to about three for the rear shoulder, which is short of what a muscle worked on purpose usually asks for. The gap is not a mistake in your training. It is just what "the same sets, split unevenly" adds up to once you write it down.
The fix costs very little. One rear-shoulder movement for two to three sets, once or twice a week, closes most of what pure rowing leaves open. It does not replace the row. It covers the part of the shoulder the row was never going to fully reach on its own.
The honest limit
Muscle-activity readings measure effort in the moment, under lab conditions, usually with a handful of trained lifters on a single machine. They are not a direct measurement of how much a muscle grows over months of real training, and a reading taken on someone else's row is not a guarantee about yours: your grip width, your torso angle, and your own shoulder shape all shift the numbers a little.
Grip strength is its own confound. If your hands give out before your back does, the set ends early for every muscle involved, and no reading taken on a machine that isolates the grip will show you that. What the research gives you is the shape of the pattern, back first, rear shoulder and biceps a real but smaller share, not a set of numbers you could reproduce on a bathroom scale. Treat it as a reason to check your own week rather than a substitute for checking it.
Adonae tracks training, food, body and watch data in one place, then uses the history to tell you which muscles are behind and whether your best lifts are moving. Free to start, works with no account. Get it for Android, or use the web app.
